THE DYNAMICS OF HEAD AND NECK IMPACT AND ITS ROLE IN INJURY PREVENTION AND

THE COMPLEX CLIN ICAL PRESENTATION OF CERVICAL SP INE INJURY

Barry S. Myers, M.D., Ph.D. and Roger Nightingale, Ph.D. lnj u ry and Orthopaedic Biomechanics Laboratory

Department of Biomedical Engineering, Department of Biological Anthropology and Anatomy

Division of Orthopaedic Surgery Duke University

ABSTRACT

This paper reviews our research on catastrophic head impact compression neck in­j u ry. On the basis of these experiments, a biomechanical model of the spine is developed in which the complex clinical presentation of cervical spine injuries may be better un­derstood. This inc ludes the significance of head rebound, head and neck decoupling, cervical spine buckling, cervical inj u ry mechanisms, basilar skull fractures, cervical in­j u ry classification, and cervical spine tolerance. Specifically, we hypothesize that impact i nj u ry shou ld be modeled as the dyna m ic response of two large masses cou pled by a seg­mented cu rved beam-column comprised of seven small masses with interposed nonlinear viscoelastic flexibi l ity elements. These impact data also provide insights into the effects of the padding on the mitigation of head and neck injury.

CATASTROPHIC CERVICAL SPINAL INJURY has remained among the most difficult and socially significant impact inj u ry problems in structural biomechanics. While these inj u ries occur through a variety of mechanisms, including direct and indirect loading, and contact and non-contact loading, head contact resulting in compression-bending neck loading remains among the most common mechanisms of injury. The volume of l iterature which has been devoted to the characterization of cervical spinal impact injury is large, and has been recently been reviewed in deta i l (Myers and Winkelstein, 1995). Despite this collection of writing, considerable confusion remains as to the basic mechanisms which result in catastrophic cervical spinal injuries including: the effects of end condition; the relationships between head motion and inj u ry mechanisms; the effects of head, neck, and torso inertia; the role of buckling in injury, and the effects of the in itia 1 orientations of the head, neck, a nd torso relative to the im pact su rface. Although previous studies have examined these variables and provided invaluable insights into the dynamic behavior of the cervical spine, few have had the long term support necessary to obtain the meaningful sample size for statistica l ly confident conclusions.

The fa i l u re of the classical static approach to neck inju ry is best i l lustrated by efforts to classify catastrophic inj u ries. Classica l ly, i t is suggested that in order to produce a lower cervica l flexion injury, the head must be flexed, causing the entire spine to flex until a motion segment is loaded beyond its tolerance (Harris et al„ 1986). However, analysis

IRCOBI Conference - Hannover, September 1997 1 5

16

of real-world head-impact neck injuries frequently results in paradoxic observations in which head motion is not consistent with the classical thinking on injury mechanism. Consider the following case history (Figure 1) . A restrained front seat passenger in a driver side leading rollover suffers a blow to the head du ring the inverted impact of the roof to the ground. The impact resulted in a lower cervical bilateral facet dislocation, a posterior arch fracture of the atlas and a posterior arch fracture of the axis (a Hangman's fracture). In this case, a midsagittal laceration anterior to the head vertex, but posterior to the hairline, clearly defined the point of head impact. This illustrates an example of multiple noncontiguous spinal injuries in which the injury mechanisms differ (i.e. a lower cervical compression-flexion injury in association with contiguous upper cervical compression-extension injuries). This cannot be explained on the basis of a simple head motion and is not consistent with the site of head impact, which one would expect to cause the head to move in extension.

FIGURE 1 Lateral X-ray illustrating a C6-C7 bilateral facet dislocation resulting from compression-flexion loading, and posterior element fractures of the at!as and axis from compression-extension loading. Both injuries were the result of a single impact anterior to the head vertex, and illustrate the poor relationship between the motions of the head in impact and the motions and injuries of the spine.

Of equal importance to characterizing injury is the prevention of injury. 1.t is widely assumed that energy absorbing devices which are designed to reduce the severity and incidence of head injury also reduce the risk for neck injury. Such devices include helmets,

IRCOBI Confere11ce - llamiover, September 1997

ai rbags, American football tackling blocks, and padded su rfaces on vehicle interiors. Despite this belief, reduction of neck injury through the use of padding has never been demonstrated i n any experimental or epidemiological study. On the contrary, a number of studies have shown that padded surfaces have no effect on cervica l spine injury risk (Alem et a l . , 1984; Nusholtz et a l . , 1983, Pintar et al., 1990), and others have presented evidence that they can increase the risk (Hodgson and Thomas, 1980; Mertz et a l . , 1978; Myers et a l . , 1991; Torg et a l . , 1976; Yoganandan et a l . , 1986). N usholtz et al . (1983) performed a study using whole cadavers and concluded that energy absorbing materials "did not necessarily reduce the amount of energy transferred to the head, neck, and torso or the cervical damage produced" . Hodgson and Thomas (1980) conducted impact tests on whole cadavers and found that distributed impacts (arising from padded impact surfaces) tend to "grip" the head and strongly influence the mechanism of injury. A quasi-static experiment by Myers et al. (1991) showed a dramatic i ncrease in cervical spine stiffness and injury risk with increasing constraints on the motion of the head. Based on these results, they concluded that restrictions on head motion may increase the risk for cervical spine injury by not al lowing the head and neck to move out of the path of the fol lowing torso. Despite these supportive arguments, there have been no studies which have studied the effects of padding by systematical ly varying the material characteristics of ·the impact surface. Therefore, one of the goa ls of our research is to create a realistic head impact environment in which we can test the hypothesis that padded impact surfaces can increase the risk and severity of cervical spine injury by constrain ing the motion of the head.

Clearly, if we are to become more effective in preventing cervical spine injuries, a cogent u nderstanding of how they occur must first be achieved. Based on our experience, we hypothesize that the cervical impact behavior may be characterized as the impulsive response of a curved segmented beam column comprised of seven smal l masses, coupled by nonlinear viscoelastic elements, bounded between two large masses with significant rotary i nertia. Fortunately, over the last fifteen years, a renewed interest in determining the basic biomechanics of neck inju ry has developed in the United States from both the Centers for Disease Control, and the National Highway Traffic Safety Admin istration a l lowing us the opportunity to test our hypothesis. lt is therefore, the purpose of this paper to review our research on catastrophic head impact neck injury which support this premise (Myers et a l„ 1997; N ightingale et a l . , 1996a, 1996b, 1997a, 1997b), to provide a biomechanical framework upon which the mechanisms of cervical spine impact injuries occur, and to discuss a framework upon which injury prevention strategies can be developed a nd refined.

MATERIALS AND M ETHODS

Experimental Apparatus

An experimental apparatus was designed to model cervical spine injury resulting from vertical head impact with a following torso (Figure 2). A steel carriage was mounted to a drop track using two linear bearing sliders and was weighted to simu late an effective torso mass of 16 kg. The value for the torso mass is based on the G EBOD output for the 50th percentile male upper torso and is an estimate of the portion of the total body mass which acts on the neck during dynamic loading. The specimen preparations were mounted to the carriage in an inverted position.

The impact surface was a 4 cm thick steel anvil with a diameter of 15.25 cm. Variation in impact angle about the y-axis (that axis normal to the sagittal plane) was achieved by mounting the anvil on a locking clevis. The impact angle was varied between -15° (posterior head impact) and +30° (anterior head impact) , accordi ng to the sign convention shown in Figure 1. The anvil was covered with 3 mm of Teflon sheet to simulate impacts onto a rigid, frictionless surface. Impacts onto a padded surface were

IRCOBI Conference - Hannover, September 1997 17

18

16 Channel Digital Data Aquleition System

FIGU RE 2 A diagram of the test apparatus showing the accelerometer on the torso mass {A), the optical velocity sensor {B), the carriage and torso mass (C), the six-axis load cell at Tl (D), the head accelerometers (E), and the anvil and three-axis load cell (F).

simulated by attaching foams to the anvil with duct tape. The Teflon covered steel surface simulated the unconstrained head end condition in 10 tests. A more constrai ned head end condition was simulated in the remaining tests (n=12) using either an expanded polystyrene foam (EPS) (E =2096.1 kPa, uy =206.2 kPa, p =0.0284 g/cm3) or a less stiff, open cell polyurethane foam (OPU) (E =158.6 kPa, uy=7.0 kPa, p =0.0277 g/cm3).

Multiaxis transduction was used to ful ly quantify the forces and moments acting on the head and neck during the impact event. Head impact forces were quantified using a Kistler 9067 three-axis piezoelectric load cell mounted u nder the impact surface. A GSE Model 6607-00 six-axis load cell mounted to the specimen was used to measure forces and moments at Tl. A PCB 302A02 uniaxial accelerometer measured torso deceleration. Sagittal plane kinematics were quantified using two PCB 306A06 accelerometers which were mounted to the head of the specimen. Impact velocity was recorded using an MTS optical sensor. The sixteen channels of transducer data were sampled at 62.5 kHz using a PC-based acquisition system. Each test was also imaged using a Kodak Ektapro EM-2 digital camera at 1000 frames per second.

Specimen Preparation

Unembalmed human heads with intact spines were obtained shortly after death. Al l specimen handl ing was performed in compliance with CDC guidelines (Cavanaugh et a l . , 1990) . The specimens were sprayed with calcium buffered isotonic saline, sealed i n plastic bags, frozen and stored at -20°. All donor medical records were examined to ensure that there were no preexisting conditions, such as degenerative diseases or spinal pathologies, which could affect the structural responses of the specimens. Donor age ranged from 35 to 80 years.

The specimens were prepared for testing in a 100 percent relative humidity chamber. The muscular tissues were removed while keeping a l l the l igamentous structures intact (with the exception of the ligamentum nuchae). The specimens were transected at T3-T4 and the bottom two vertebrae were cleaned, defatted , and cast into a luminum cups with reinforced polyester resin . Care was taken that the most rostral uncast vertebra was free of resin and was al lowed full range of motion. The C7-Tl i ntervertebral disc was oriented at 25 degrees to horizontal to preserve the resting lordosis of the cervical spine (Matsushita et a l . , 1994) . Following casting, a triaxial accelerometer array was attached to exposed parietal bone using dental acrylic and bone screws. A jig was used to ensure that the array was positioned parallel to the sagittal plane. The position of the array relative to the Frankfort anatomical plane was determined radiographically using the

IRCOBJ Conference - Hannover, September 1997

auditory meati and the inferior margins of the orbits. Final ly, photographic target pins ( 4.0 mm diameter) were inserted in the anterior vertebral bodies, the spinous--processes, and lateral masses of C2-C7. The pins were used for photogrammetric analysis of the vertebral motions.

Experimental Protocol

Cadaveric specimens were inverted and mounted to the carriage of the drop track system in the anatomically neutral position. Break-away sutures were passed through the ear lobules and nasal septum and were tied to the suspension frame to maintain the neutral orientation of the cervical spine. Each specimen was raised to the desired height and the cervical spine was mechanically stabilized by manua l exercise through a flexion-extension range of 60° for fifty cycles (McElhaney et a l . , 1983). The specimens were dropped from a height of 0.53 m, which was less than that required to cause a skull fracture, yet sufficient to produce cervical spine injury (McElhaney et al. , 1979) . Following impact testing, anteroposterior and lateral radiographs were obtained and the specimen was disarticulated at 0-Cl and the head was weighed. To document the injuries, dissection was performed on both the heads and cervical spines.

Data Analysis

Al l transducer data were uploaded to a Sun SparcStation 2 for analysis. Digital filtration was performed in accordance with the Society of Automotive Engineers standard for head and neck impacts (SAE J211b Class 1000). In order to determine inertial head loading and eva l uate the risk of head injury, l inear acceleration of the center of gravity of the head was determined from the head mounted accelerometer array.

In impact testing, events other than material fa ilure can be associated with a drop in the axial force with time. These include buckling, slip of the head on the impact surface, and unloading of the neck due to rebound of the torso. The following criteria were used to define mechanical failure and relate mechanical failure to the occurrence of injury. Each decrease in force with increasing time was eva l uated. The high speed images were coregistered to the load cell data and were analyzed to define the kinematics of each motion segment at the time of decreasing load. A decrease in load was related to a specific injury if the kinematics and injury mechanism were mutually consistent at the time of the decrease in load. Decreases in load as a result of slip at the head­impact surface were excluded by examination of image and head acceleration data. Decreases in load associated with increases in length of the neck were considered to be mechanical unloading and were simi larly excluded. Any decrease in load with a concomitant increase in bending moment which also demonstrated a rapid transition from one mode of deformation to another mode of deformation on high speed images, and no evidence of injury at the site of the transition was defined as a stable buckle.

The cl inical stability of a cervical spine injury is a measure of its severity. Stable injuries genera lly do not result in a progressive neurological deficit and are treated con­servatively. Unstable injuries are much more l ikely to have neurological involvement and are treated by open or closed reduction and fixation. The stabi lity of all the injuries produced in the drop tests was assessed using two methods. First, the injured motion segments were manipulated and obvious gross motions were defined as unstable. Second, the injuries were evaluated using the "one column plus one element" stability criteria outlined by Panjabi et al. (1979). lf all the elements of the anterior column (anterior longitudinal ligament, disc, and posterior longitudinal ligament) and any element of the posterior column (facets, ligamentum flavum, interspinous ligament, and supraspinous l igament) are disrupted, then the injury is considered to be potentially unstable. S imi­larly, the injury is classified as unstable if the posterior column and any element of the anterior column are disrupted.

IflCOBI Conference - Hannover, September 1997 1 9

The im pulse of the compressive component of force at Tl was calcu lated for a l l the impacts by integrating the axial force history. Differences in axial impulse between the padded and rigid tests were evaluated using two-way ANOVA. The effect of padding on the frequency of injury was examined using a x2 comparison of proportions for two independent samples with a continu ity correction. The effects of padding on the peak head and neck forces were examined using t-tests.

RESULTS

Using this test system , a total of 22 impact tests have been performed to date, producing basilar sku l l fractures and cervical spine injuries in 16 cases �Tables 1 and 2) . Cervical spine injuries included anterior disc tears, anterior longitu inal ligament ruptures, Jefferson fractures, Hangman's fractures, odontoid fractures, burst fractures, facet dislocations, and posterior element fractures.

TABLE 1: Subject Data and Drop Test Results Res.0 Axialb Res.c Head Neck Neck

Age, Vel. Angle Force Force Force Timed Impulse Lag TEST Sex {m/s) {deg.) {N) {N) {N) (msec) {N · s) {msec) HIC

Rigid surface N05-R+30 36,M 3.23 Ant. (+30) 8790 1552 1593 8.3 35.9 1.8 497 N18-R+15 -,M 3.26 Ant. ( +15) 7498 1863 1895 6.4 62.6 6.4 1935 D41-R+15 69,M 3.11 Ant. ( +15) 8604 NI NI NI 56.6 2.0 132-R+15 78,M 3.18 Ant. ( +15) 8234 2416 2612 3.9 38.9 1.9 1361 N26-R+O 65,M 2.43 Vertex (0) 7638 NI NI NI 47.7 1.5 N24-R+O 62,M 3.20 Vertex �O) 8566 1839 1973 2.2 40.7 2.2 N22-R+O 71,M 3.26 Vertex 0) 8111 1955 2120 6.5 46.9 6.5 490 Nll-R-15 55,M 3.14 Post. (-15) 11621 NI NI NI 24.1 2.2 543 N13-R-15 35,F 3.28 Post. (-15} 5615 NI NI NI 20.6 1.3 704 UK3-R-15 62,M 3.13 Post. (-15} 5093 NI NI NI 30.7 1.0 1783

Padded surface N21-P+30 61,M 3.13 Ant. ( +30) 1760 1632 1662 14.8 42.7 5.3 50 N23A-P+30 46,M 3.03 Ant. (+30} 3608 NI NI NI 39.7 5.8 77 N23B-P+30 46,M 3.51 Ant. (+30) 3857 2240 1698 18.7 31.7 7.5 197 I08-P+15 80,M 3.15 Ant. ( +15) 5946 2915 2918 30.5 78.6 4.2 110 111-P+15 63,F 3.20 Ant. (+15) 3115 967 972 14.0 71.9 4.6 118 104-P+15 63,M 3.19 Ant. (+15) 3383 1675 1698 18.0 74.1 3.8 N03-P+O 75,M 3.08 Vertex (0) 5664 3172 3509 18.2 62.7 2.7 84 N02-P+O 75,F 3.14 Vertex �O� 3452 715 793 14.7 76.0 2.6 122 040-P+O 53,F 3.16 Vertex 0 4187 1438 1440 16.7 81.1 1.7 N19-P-15 42,F 3.07 Post. (-15) 2604 1011 1037 18.8 22.6 9.0 175 NA2-P-15 61,M 3.16 Post. (-15) 4749 1968 2091 15.6 35.1 4.2 270 125-P-15 59,M 3.07 Post. (-15) 5963 2558 2574 18.4 39.7 2.9 384

NI The specimen had no injury. a Peak resultant head force. b The magnitude of axial neck force at injury. c The magnitude of the resultant neck force at injury. d The the elapsed time between impact and injury.

The impact dynamics of the head and neck are bimodal and reflect the vibrations of a two mass system with an interposed poorly coupled viscoelastic spring striking an impact surface (Figure 3). For the rigid impacts, Mode 1 is attributed almest entirely

20 IRCOBI Conference - l/annover, September 1997

TABLE 2: lnjury Results Class Head

TEST Pathology (Allen, 1982) Stability Motion Rigid surface

N05-R+30 C3 burst fx., vc unstable extension C3-C4 disc and ALL, DE stable C4-C5 ALL DE stable

Nl8-R+l5 Cl lateral mass fracture, vc stable extension C2 hangman's, C2-3 disc and ALL, DE unstable C6-7 bilat. facet dislocation DF unstable

D41-R+l5 None extension 132-R+l5 C5-6 disc, L capsular lig., ALL DE stable extension N26-R+O None extension N24-R+O Cl 2 part posterior arch fracture, CE stable flexion

C2 hangman's fracture DE stable N22-R+O Cl 3 part comminuted fracture vc unstable extension Nll-R-15 None flexion N13-R-15 None flexion UK3-R-15 None flexion

Padded surface N21-P+30 Cl anterior ring fx., CF stable extension

C4 spinous process fx., CE stable C5 spinous process fx., CE stable C5-C6 disc, left capsular lig. and ALL DE stable

N23A-P+30 None extension N23B-P+30 Cl 2 part right aspect fx., vc stable extension

C3-C4 disc and ALL, DE stable C4-C5 disc and ALL DE stable

108-P+l5 C2 hangman's fracture and burst DE unstable none 111-P+l5 C2 type III dens + comminution, unstable none

C4 body, R lamina vc stable 104-P+l5 Cl 2 part posterior arch fracture, CE stable none

C2 hangman's, C2-3 disc, ALL, DE unstable C7-Tl posterior ligs. DF stable

N03-P+O C4-5 capsular lig., stable flexion CS-6 disc, DE stable C6-7 bilat. facet dislocation DF unstable

N02-P+O Cl ant. ring, CF stable none C2 hangman's + type I I I dens, DE unstable C6 R lamina and pedicle, vc stable C7 burst vc unstable

D40-P+O Cl 3 part comminuted fracture, vc unstable none C3-4 disc, ALL, spinous proc., DE stable C5 {burst), C5-6 PLL, vc unstable C6 R lamina and pedicle vc stable

Nl9-P-15 C2-3 disc, ALL, C2 ant. avul. fracture, DE stable flexion C3-4 disc, ALL C3 ant. avul. fracture DE stable

NA2-P-15 C3-4 disc, ALL, L capsular lig., DE stable flexion C5-6 disc, ALL DE stable

125-P-15 Cl-2 L capsular lig., stable flexion C3-4 disc, ALL, PLL, L capsular lig. DE unstable

• Upper cervical spine injuries were classified using a system similar to that of Allen et al. (1982)

IRCOBI Conference - Hannover, September 1997 21

to stopping the head and had a duration of 4.3±1.6 mil l iseconds. During the first half of this head inertial loading mode, the head impact force reached a maximum with no concomitant neck force (Figure 3). Neck loading at Tl was not observed until the latter ha lf of Mode 1. For the padded impacts, the head contact times during Mode 1 were significantly increased. Therefore, the first mode contained loading by the torso in addition to the force requ i red to stop the head. The i nertia and compliance of the head mass caused rebound loading. Head rebound forces on the order of 10 to 35% of the peak head force were commonly observed, indicating that in impact, the neck must manage both the momentum of the torso a nd the head (Figure 4). Mode 1 durations for the padded impacts could not be cak:u lated because the increased coupling of the head a nd cervical spine resulted in less separation between modes. For both impact surfaces, Mode 2 represents loading of the impactor surface by the head, cervical spine, and the effective torso mass. The duration of this neck impact surface loading mode for the rigid impact was 27.3±14.3 msec. In all the tests there was a delay in the onset of measured neck load with respect to the head load. This lag in response at Tl was 1 .6±0.3 msec for the rigid impacts which was significantly different than the 4.7±1.3 msec for the padded impacts (p<0.001, Table 1) . The lag is evidence that the head and cervical spine are not coupled du ring the first half of the head impact mode.

..,. c � „ � g u.

N18 HIC.1939, t,•0.0046, !,•0.0064 8000 '!!: :·1:1·;,::ii:· 400

350 6000 300

250 4000 200 :Il 150 �

2000 100 c � 50 0 0 >

-2000 � -4000

§ -6000

Time (seconds)

„ c 0 l „ � g u.

040 HIC-89.6, t,•0.0191, !,20.0379 2000 ,., 1 , „, 100

1000 ::! !:::[; " ,. !

. , -1000 t-1· .,...,.,.,��"""""�__,....�__,..__...,.. -2000 „ 1: 1• '

: f : , -3000 f-+_ ! -;;-. +""'S'444+wi'.:.Ml'+4"""it.:.:...L----1

Time (seconds)

FIGURE 3 Magnitude of the head and neck axial forces, and the head center of gravity acceleration for an impact into a rigid surface oriented at +15° (anterior) (left) and into a padded surface oriented at +0° (right). A bimodal response consisting of a head inertial deceleration mode (Mode 1), and a neck impact surface loading mode (Mode 2) are seen in both the padded impact and the unpadded impact. Analysis of the rigid impact (left) kinematic data revealed a stable buckle followed by a C6-C7 bilateral facet dislocat1on, and a basilar skull fracture. Addition of padding significantly lowered the head accelerations, HIC, and the Mode 1 head impact forces, however the bimodal response was still observed. Addition of padding also increased the head contact time in the impact surface. lnteresting, the basilar skull fracture occurred during the neck impact surface loading mode (Mode 2).

Head motion was not related to injury mechanism. Cervical spine inJunes were produced withi n 9 msec following head contact with a rigid surface and in less than 20 msec in al l but one of the padded surface impacts. In contrast, significant head motions, as characterized by rotations greater than 20°, did not occur until between 20 msec and 100 msec fol lowing head impact (larger times in the padded impacts a nd shorter times i n the rigid impacts). During the initial phases of impact, the heads were observed to u ndergo smal l flexion or extension motions which were not in the di rection of the final motion owing to the curvature of the head. The final head motions, the kind which would be observable on a 60 Hz record ing or to an observer, were directly related to the point of impact. Specifical ly, for head impacts forward of 15°, the final head motion was extension rotation. For impact surfaces oriented at 15° and posterior, the final head motion was flexion. Analysis of the injuries and the mechanism of injury as determined by local spinal kinematics, and force analysis was unrelated to the final head

22 IRCOBI Conference - Hannover, September 1997

Direction of head acceleration ((·)inferior,(+) superior).

J -�il!L f '�J;�'.z 0: '. d 1.0 ....-�....--.---,--.----.-�---,r---..--......--, 100 �

\ '\ „ .... o.o A

z ::.. ·1.0

Cl> e 0 LL ·2.0

·3.0

-4.0 o.o 10.0

/ . . ; ' ,\ ,\ '\ • J •J

20.0 30.0 40.0 Time (milliseconds)

o.o

-100

·200

-300

50.0 -400 60.0

� ::::1 e: a. Cl> a :::c Cl> [ )> 0 p g

FIGURE 4 Vertical force and acceleration history for the head and neck. lnitially, the head hits the contact surface, large head forces accelerate the head up (negative ac­celerations). After C, the vertical force of the neck driving the head back down into the impact surface exceeds the head-impact surface force, and the head is driven down (positive acceleration) into the impactor by the neck. In that regard, the neck loading reflects inertial contributions of the head and the torso.

motions (Table 2) . lnjuries produced prior to the head flexion motion i ncluded eight extension injuries, one compression injury, and one flexion injury. lnjuries produced prior to the head extension motion included, n ine extension injuries, four compression i n juries and two flexion injuries. Of those impacts not resu lting in significant head motions, compression, extension and flexion injuries were produced. Further, head motion was unable to explain the occurrence of mu ltiple noncontiguous injuries in which differing i njury mechanisms were observed .

Oynamic buckling of the cervical spine following head impact was observed in each impact within a 3 to 8 msec interval (Figure 3) . The buckle was characterized by a rapid transition from a compression mode of deformation to a bending mode of deformation (Figure 5) . Because the buckle was observed in specimens without injury, and because the i njured spines demonstrated post-bucking stability prior to injury, we concluded that the buckle did not result in material failure. Oeforming as a fixed-pinned precurved member, the spine underwent a characteristic postbuckled configuration with m idcervical extension and lowermost cervical flexion, while the upper cervical spine flexed, extended, or remained a l igned and compressed. Un l ike head motion, the post-buckled deformation and resultant force location are able to explain the mechanism and patterns of i n jury (Figure 6) .

lnterestingly, three basi lar skull ring fractures were detected following craniotomy and removal of the brain (N18, 040, and 111) . No cranial vault fractures were detected. In each of the three cases, cervical injuries were also detected. Two cases (N18, a nd 040) i nvolved contiguous fractures of Cl, while the third (111) had noncontiguous cervical i njuries. Each basilar skull ring fracture originated in the dorsum sellae or the occipital clivus a nteriorly. The fractures propagated bilaterally through the sphenoid bone, tempo­ral bone, or adjoin ing sutures. The fractures then propagated posteriorly around, or into, the foramen magnum forming either complete (040 and N18) or incomplete ring frac­tures (111) . The rostral-most (anterior) portion of each fracture was displaced into the crania l vault by 3 to 5 mm. The caudal-most (posterior) portion of each fracture showed less displacement. Each fracture was mechanically stable to pa lpation as a result of soft tissue attachments, incomplete fractures, and interdigitation of the fracture fragments. With effort, the basilar sku l l fragment could be displaced into the vault. Because of the mechanical stabi l i ty of the fracture and minimal displacement of the posterior portions of the fractures, the injuries were not read ily apparent on external examination of the skul l following craniocervical dislocation, nor were they detected on plain films.

IRCOBI Conference - Hannover, September 1997 23

FIGURE 5 High speed digital images of a rigid impact at -15° illustrating buckling effects (the images have been modified to exclude facial features). The ima�es show the normal lordotic curve (2 msec), followed by a transient higher order mode t3 msec) which decays quickly ( 4 msec) to a first order buckle (5 msec). The buckle is then stable (non-injurious) and results in a characteristic post buckled deformation including regions of flexion, extension, and compression within the cervical spine.

LmcorAC'lic:m o( Lood Voe10r

FIGURE 6 Schematic illustration of the postbuckled spine, showing a representative location of the resultant force and the injury distribution. Unlike head motion, the post­buckled deformation and resultant force location are able to explain the mechanism and patterns of injury. (CF= compressive flexion, CE= compressive extension, VC = vertical compression, DE = distractive extension, DF = distractive flexion)

Each of the basilar skull fractures occurred during the neck-loading mode (Mode 2) of the impact (Figure 3). Specifica lly, using the defined injury criteria a nd combining the kinematic a nd kinetic data sets, basilar skull fractures occurred 15.4±3.2 msec following peak impactor force. Head lnjury Criteria for the rigid and two padded basilar sku l l fractures were 1935, 90, and 119, respectively. Time corridors (Tl and T2) selected

24 IRCOBI Confere11ce - Hannover, September 1997

to maximize the HIC calculation fel l within the head inertial loading mode (Mode 1) for each of the three injuries. Thus, the time range used in the HIC calculation did not include the time at which the sku l l fractures occurred. The axial (spatial ly fixed , superior-inferior direction) neck forces to produce the fractures were -2494, -2126, and -2085 N for the rigid and padded injuries (mean = -2235±225 N). The axial forces requ i red to produce basilar sku l l fracture were larger than the forces requ i red to produce the concurrent cervical injuries in each of the three cases (Table 2).

Padding was found to significantly increase the risk for injury and to increase the severity of injury. Using the groups of data in which both rigid and padded impacts were performed, we found that a l l n ine of the padded impacts and four of the nine rigid impacts produced cervical spine injury (Table 3). Performing an ANOVA on these data, we found that the frequency of injury in the padded impacts was significantly greater than the frequency of injury in the rigid impacts (p=0.0375). In the same context, only two of the rigid impacts were found to be clinically unstable, while seven of the n ine padded impacts were unstable.

TABLE 3: l njury Frequency and Severity

Rigid Impacts lnjured Unstable

Anterior 2/3 Vertex 2/3

Posterior 0/3

Total 4/9

1/3 1/3 0/3

2/9

Padded Impacts lnjured Unstable

3/3 3/3 3/3

9/9

3/3 3/3 1/3

7/9

For the rigid impacts, the axial impulses were 41.0±14.2 N-s. For the padded im­pacts, the axial impulses were 60.2±21 .9 N-s. The axial impulses were grouped based on impact angle and impact surface in order to perform a two-way analysis of variance. The impu lses in the padded impacts were sign ificantly larger than the impu lses in the rigid impacts (p=0.00023). Two-way ANOVA a lso found significant differences between axial impu lses as a function of impact a ngle (p<0.0001) with the posterior impacts pro­duced the smal lest impulses and the anterior and vertex impacts produced larger impulses (Table 1) .

Padding sign ificantly lowered the risk for head injury. Peak resultant head force applied by the impact surface was significantly lower for the padded impacts, -4127±1375 N, than for the rigid impacts, -8297±1572 N (p=0.02). Similarly, the average HIC for the padded impacts, 136±32, was significantly lower than the average H IC for the rigid surface, 1010±534. Average axial neck force to produce the first neck injury was -1948±666 N and did not vary with impact surface.

DISCUSSION

Catastrophic head impact neck i njury remains a significant societal problem. As a direct result, a large volume of literature has been written on neck injury in which static relationships between head motion and neck injury are hypothesized . U nfortunately, as in the case presented herein , real world cases often present with findings which cannot be explained using this static approach. This suggests that neck i njury is the resu lt of the dynamic responses of a more complex mechanical system. With this in mind, the purpose of this investigation is to understand the dynamics of head impact neck injury, to create a mechanical paradigm in which the responses of the neck are better understood, and to a l low a rational investigation of strategies for neck injury prevention.

IRCOBI Conference - Hannover, September 1997 25

Central to the testing of any hypothesis is the demonstration of the impact model's valid ity. Using our experimental apparatus, we produced a variety of clinically observed injuries, including Jefferson fractures, Hangman's fractures, burst fractures, posterior element fractures, bilateral facet dislocations, and basilar sku l l fractures. The injuries in this study were produced in the absence of preflexion (i.e. the normal lordosis of the spine was preserved) . Additional ly, the distribution of injunes P.roduced is consistent with the distribution of inju ries reported in epidemiology studies (Myers and Winkelstein, 1995).

These data i l l ustrate the ease with which the neck may be injured, and i l lustrate how neck i nj uries can occur with the pelvis properly and fully restrained owing only to the mass of the u ppermost portions of the torso. Six of the specimens suffered no injuries whi le 16 were injured using an impact velocity of 3.1 m/s. This was consistent with the measurements by McElhaney et a l . (1979) in which a head impact velocity of 3.1 m/s was thought to be a critical velocity for neck injury. lmportantly, our model used an effective dynamic torso mass of only 16 kg, which represents only approximately 40% of the isolated torso mass of a 50th percentile male.

In considering only head and neck injury potential , the impact may be regarded as a one body problem involving the deformation of a neck spring beneath an effective dynamic torso mass. The lag between the neck load and the head load indicates that the torso a nd neck are poorly coupled. Therefore, the spring should be nonlinear, show an initial low stiffness region , and stiffen with increasing deformation (Myers et a l . , 1991). With this model , i nj u ry potential exists when the head is constrained and the neck is cal led upon to stop the moving torso. l nterestingly however, two of the impacts to rigid, low friction surfaces resulted in catastrophic spinal injuries. Thus, injury was produced in the complete absence of constraints imposed by the contact surface. This differs from static studies which were unable to produce compression neck injuries in the absence of a constrained head end condition (Myers et a l . , 1991, Roaf, 1960), a nd is explained by the i nertial loads imposed by the head. Specifical ly, these experiments show that in addition to managing the momentum of the torso, the neck must also accelerate the head out of the path of the torso. They also show that head rebound from the impact surface creates a significant portion of the neck load. In that regard, neck injury must be modeled as a two body problem in which neck loads are generated by the torso mass, the impact surface and the head mass.

As i n prior studies, head motion was unrelated to injury mechanism (Nusholtz et a l . , 1983). This occurred in part because the head motion lags the applied forces which produce i nj ury, and does not occur until weil after neck injury. Also not explained by head motion, is the frequent occurrence of multiple noncontiguous injuries of differing mechanisms at different levels within the spine. This was seen in the case study (Figure 1) , and also seen in the experimental data, i ncluding specimen N 18, whose injury distri­bution and point of impact are identical to those presented in the case study. Returning to the mechan ical model of the segmented beam column helps to explain these other­wise paradoxical observations. These experiments show that the spine acts as a slender, curved beam column with a fixed base and pinned apex. Once a critical load is reached, this structure snap-through buckles with a rapid transition from a predominantly com­pression mode of deformation to a predominantly bending mode of deformation (Figure 6) . As in prior static studies (Myers et a l . , 1991 ) , the cervical spine shows post-buckling stabil ity, as it can be buckled without injury. Thus, buckling is a structural fa i lure and not a material fa i lure (a neck injury). While buckling is not injury, the post-buckled mode shape does explain the mechanism of these injuries. The buckled deformation exhibits regions of extension in the middle cervical spine a nd flexion in the lower cer­vical spine. As a result, the neck load vector passes behind the middle cervical spine and causes compression-extension, and in front of the lower cervical spine and causes lower cervical compression-flexion . Thus, by treating the spine as a seven segment beam column which is sufficiently slender to buckle, interposed between the torso and head masses, the relationships between the point of head impact and the type of neck injuries produced becomes weil defined.

26 JRCOBI Conference - Hdltnover, September 1997

Whi le able to explain the injuries seen in this study, the first order buckle does not explain the production of midcervical (C3 to C6) flexion injuries. Yet clinical studies show that this is a common injury. Additional ly, previous research has demonstrated that preflexion may be one mechanism for producing midcervical compression-flexion injuries (McElhaney et a l . , 1983, Pintar et a l . , 1990, Torg et a l . , 1990). Another, mechanism of midcervical flexion injuries may be the development of higher order buckling modes (Fig­ure 5) . In these experiments, the higher order mode shape, which produces midcervical flexion decayed prior to injury. Using a numerical model validated against these experi­ments ( N ightingale et a l . , (submitted) ; Camacho et a l . , (submitted)) , showed that the buckled mode shape is most dependent on the flexion-extension flexibi l ity of the spine, a nd that this higher order mode is expressed because of the inertial forces developed in each of the vertebral masses during the snap-though motion (Figure 7) . lt is possible that at higher impact velocities, in which the torso loads the neck more rapidly, injury

· may occur during this higher order mode, and result in midcervical compression flexion inJuries.

FIGURE 7 Posterolateral (left ) views of the cervical vertebrae reconstructed from axial CT images (Camacho et at , 1998). From these meshes, the centroid position, mass, and sagittal plane moment of inertia were calculated for each vertebra. The Hybrid finite element head/multibody dynamics neck model (right) was derived from the reconstructed mesh and spinal flexibility data (Nightingale et al. submitted). A rigid face was coupled to the rigid maxilofacial region of the skull and assigned mass properties such that the mass, moment of inertia, and centroid of the entire head matched the data of Walker et al. (1973). Validated against the drop test data, this model shows that the expression of the higher order buckling mode is the result of inertial forces from the small masses of each vertebrae which oppose the expression of the first mode buckle.

lt well recognized that the spine is a complex geometric structure with nonlinear, rate dependent responses. The results of our studies suggest the need for a more com­plex mechanical paradigm. That is, by considering the effective dynamic torso mass, the head mass, the· nonl inear curved slender segmented beam column and each of the vertebral masses, the dynamic responses of the spine can be more fully understood. Ad­ditional ly, by organizing the neck injury literature according to this paradigm, a coherent u nderstanding of neck injury mechanisms and injury classification emerges. Specifically, at the time of i njury, a resultant force with some eccentricity acts on the spine at the

IRCOBI Conference - Hannover, September 1997 27

TABLE 4: Cervical Spine lnjuries: A Classification Based on Applied Forces with Experimental Validation

Compression

Jefferson fracture Mu ltipart atlas fracture Vertebral body compression fracture Teardrop fracture

Compression-flexion

Teardrop fracture Burst fracture Wedge compression fracture Hyperflexion spra in Bi lateral facet dislocation Unilateral facet dislocation

Compression-extension

Hangman's fracture Clay-shoveler's fracture Posterior element fractu re Anterior longitudinal ligamentous rupture Anterior disc rupture Horizontal vertebral body fracture Teardrop fracture

Tension Occipitoatlantal dislocation

T ension-extension Hangman 's fracture Anterior longitudinal ligamentous rupture Disc rupture Horizonta l fracture of the vertebral body Teardrop fracture

Tension-flexion Bi lateral facet dislocation Un ilateral facet dislocation

Torsion Atlantoaxia l rotary dislocation Unilateral atlantoaxial facet dislocation

Shear Odontoid fracture Transverse ligament ru pture

Lateral Bending (in combined loading) Asymmetrie injury Nerve root avulsion Peripheral nerve injury

point of injury. By considering this load as a force and an associated couple, a clear relationship between neck resultant force and injury is produced (Table 4).

lnterestingly, three ring type basilar skul l fractures were observed among 16 speci­mens with injuries, an incidence of 18.7%. This suggests that basilar sku l l fractures are a

28 IRCOBI Conference - Hannover, September 1997

common consequence of near vertex head impact. In each case, the skull fractures were accompanied by cervical injuries and the skul l fractures occurred at forces slightly larger than those required to produce the cervical injuries in the same specimens. Yoganan­dan et al. (1986) reported four skull fractures from head impacts. Three of these had concurrent cervical injuries, including the one specimen with a true basilar sku l l fracture (i.e. without cranial vault injuries). Alem et a l . (1984) reported a case of a basilar sku l l fracture as a result of head impact, and d id not detect other cervical injuries. These results suggest that the sku l l base may have a slightly larger force tolerance than the cervical spine. As a result, compression mediated basilar skul l fracture wil l more often than not be associated with cervical inju ry.

Alem et a l . (1984) was among the first to report basilar skul l fracture as a result of near vertex head impact. Based on their observations, it remained unclear if basi lar sku l l fractures were the result of head impact acceleration or neck loading. By measuring neck and impact surface forces, our study demonstrates that basi lar skull fractures occur toward the end of the dynamic event in the neck impact surface loading mode. At this time, the head is compressed between the cervical spine and the impact surface, and head inertia is smal l . In that regard, basilar sku l l fracture as a result of near vertex head impact represents a quasi-static fai lure mechanism which is unrelated to the head acceleration. lt is also, therefore, unrelated to head acceleration based injury criterion l ike the HIC. In this context, basilar skull fractures may be thought of mechanistica l ly as a fracture of the Cl motion segment of the cervical spine. This has implications for injury prevention. Specifically, while head injuries are readily m itigated by the addition of surface padding (as measured by a decrease in H IC), neck injuries are insensitive to, and perhaps potentiated by the addition of surface padding. Thus, it is unl ikely that addition of padding to an impact surface wil l prevent basilar sku l l fracture as a result of near vertex head impact. Additional ly, basilar sku l l fracture can occur at drop heights of 0.53 m, a height at which skull fractures are uncommon, however, cervical injuries are common. lt should be recognized that these fractures occur through other mechanisms however, and that these other mechanisms may be governed by other tolerance criterion.

The effects of padding on injury risk were examined in this study by simulating extremes in impact surface properties from rigid, low friction, unconstrained surface to a highly deformable, pocketing surface. The materials used i n this experiment are more compliant than those paddings and l iners that are currently in use by the automotive and other safety industries. Ana lysis of the injury data shows that there was a significantly greater frequency of injury in the padded impacts than in the rigid impacts (p=0.0375). This shows that pads do not protect the neck and supports the hypothesis that padded surfaces may increase the risk of cervical spine inju ry. The effect of padding on injury risk was particularly apparent in the posterior impacts: none of the rigid impacts produced i njury, and a l l of the padded impacts produced injury. In the padded posterior impacts, the head and neck have an initial component of velocity in the anterior di rection relative to the impact surface. The anterior components of forces applied to the head by the impact surface and by the neck, increase the escape velocity. These were sufficient to move the head out of the path of the following torso before i njurious loads could develop in the neck. However, in the experiments where a padded surface was used, the deformed pad applied posteriorly di rected forces which decreased the anterior head escape velocity. As a resu lt, the neck was subjected to significantly more of the torso momentum (p=0.00023); that is, the neck was forced to stop the moving torso, and the neck susta ined more injuries which tended to be more severe. lmportantly, the surface padding used in this study was able to significantly reduce the magnitude of the measured head impact force and therefore the risk for head injury. H IC values were reduced from an average of 1045 to an average of 159. Thus, this padding surface protected the head and did not protect the neck. Therefore, these experiments suggest that highly deformable padded contact surfaces should be employed carefu l ly in environments where there is the risk for cervical spine injury. In addition, the results suggest that the orientation of the head, neck, and torso relative to the impact surface is of equal if not greater importance

IRCOBI Conference - Hannover, September 1997 29

in neck injury risk. Additional research and a large number of specimens is necessary to identify which head and neck positions and impact angles are most l ikely to produce injury, a nd which padding materials and material thickness provide the optimal balance between head protection and neck protection. Given the prohibitive cost of cadaver studies, perhaps these questions will be best answered using a computational model.

The results of our impact studies, together with the works of others comprise almost 50 human head and neck studies in which adequate instrumentation are available to characterize the neck loads at the time of injury (Yoganandan et a l„ 1986, Pintar et a l . , 1995). In that regard, they serve as a database for the formulation of a human cervical tolerance. Among the important variables for defining this tolerance are gender and the presence of preflexion. That is, the male cervical spine is stronger than the female spine (Nightingale et a l . , 1997b) , a nd preflexion, which minimizes bending stresses, requ i res larger forces to fa i l u re than a neutrally positioned spine. Also important, thought not well characterized, is the sign ificance of age related changes and the possible gender differences of these changes. Not surprisingly, the methods by which these variables are accounted for can influence the estimated tolerance. Combining data sets with preflexed and neutral spines as well as men and women donors, we have suggested a resultant force tolerance to near vertex head impact of 2.75 to 3.44 kN (Myers and Winkelstein , 1995) . More recent ana lyses of these data using only males, both preflexed and neutral, suggested a tolerance of the young adult male between 3.58 to 3.73 kN (Nightingale et a l . , 1997b) .

CONCLUSIONS

1 . Neck injuries are commonly produced at velocities of 3 .1 m/s with an effective dynamic mass of only 40% of the total torso mass.

2. Head and neck dynamics include a bimodal response of a head impact mode, and a neck impact surface loading mode. Characteristic features of this response include the poor coupl ing of the head and neck owing to an initial low neck stiffness, and head rebound.

3. Head motion is unrelated to the neck inju ry mechanism as neck injuries occur prior to significant head motions.

4. The cervical spine buckles as a result of head impact. The buckle is not an inju ry, however, the post-buckled mode shape is able to expla in the distribution of neck injuries, a nd the relationship between head forces and neck injuries.

5. Higher order buckling modes can be observed in the spine and are the result of inertial loads in each of the vertebral bodies.

6. Basilar sku l l fractures are a common, difficult to detect, consequence of compression neck load ing. The inju ry behaves mechanistically l ike a neck injury as it occurs during the neck impact surface loading mode, and it is described by a neck load measurement and not a head acceleration criterion.

7. Cervical spine injuries can be classified by examining the location of the resultant force on the motion segment.

8. Constraint of head motion due to pocketing in a highly deformable padded surface may increase the risk of cervical spine injury; however, the orientations of the head, neck, and torso relative to the impact surface are of equal if not greater importance. These pads sign ificantly reduce peak head forces and therefore reduce head injury risk. Determination of an optimum pad is the subject of ongoing investigations.

30 IRCOBI Confere11ce - Jla1111over, September 1997

9. Sufficient data are available for the formulation of a meaningful near vertex head impact tolerance for the neck; however, the tolerance is a function of a number of different variables, and several methods exist to combine these data into a n average human tolerance. One such ana lysis suggests a resultant tolerance of 3.58 to 3.73 kN for the young adult male.

10. Head impact neck i njury should be modeled as the dynamic response of two large masses coupled by a segmented curved beam-column comprised of seven smal l masses with interposed nonlinear viscoelastic flexibil ity elements. Though complex, this approach creates a cogent relationship between head impact, head motion, buck­l ing dynamics, i njury classification, i nj ury mechanism, and injury prevention and is able to expla in the often confounding clinical presentation of cervical spinal i njuries.

ACKNOWLEDGMENTS

Research in the lnjury and Orthopaedic Biomechanics Laboratory has been provided by the the Department of Transportation N HTSA Grant, DTNH22-94-Y-07133, the De­partment of Health a nd Human Services CDC Grant, R49/CCR402396-10, the National Science Foundation, and the National Institutes of Health, MSTP program. The views and opinions expressed in this paper are those of the author and do not necessarily represent those of the funding organizations. A large number of dedicated and capable investigators have made contributions to the research presented herein. The most long­standing and significant is that of Dr. James McElhaney. 5ignificant contributions have been made by Drs. Wil l iam Richardson, Linda Gray, Tom Best, Brian Doherty, Robert Hopper, and Jacqueline Paver. Significant contributions have also been made by Chris Van Ee, Danny Camacho, Beth Winkelstein , Todd Woolley, Judge Robinette, a nd James Yu.

REFERENCES

Alem, N .M . , N usholtz, G .S . , Melvin , J .W. , "Head and neck response to axial impacts" . Proceedings of the 28th Stapp Car Crash Conference, 5AE Paper 841667, 927-940 1984.

Camacho, D.L .A. , N ightingale, R.W., Robinette, J . J . , Vanguri, 5 .K . , Coates, D .J . , My­ers, 8 .5 . , "Flexibi lity measurements for and validation of a computational head-neck model for near-vertex head impact" . Proceedings of the 41th Stapp Car Crash Confer­ence, (submitted) .

Camacho, D.L .A. , Hopper, R .H . , Myers, 8.5. , "An improved method for finite element mesh generation of geometrically complex structures with application to the skul lbase" . Journal of Biomechanics, i n press, 1998.

Cavanaugh, J .M . , K ing, A . I . , "Control of transmission of HIV and other bloodborne pathogens in biomechanical cadaveric testing" . Journal of Orthopaedic Research, 8: 159-166, 1990.

Harris, J .H . , Eideken-Monroe, B., and Kopaniky, D.R. , "A practical classification of acute cervical spine injuries" . Orthopaedic Clinics of North America, 17:15-30, 1986.

Hodgson, V.R., Thomas, L .M . , "Mechanisms of cervical spine injury during impact to the protected head" . Proceedings of the 24th Stapp Car Crash Conference, SAE Paper #801300, 17-42, 1980.

Matsuchita, T., 5ato, T.B . , Hirabayash i , K . , Fuj imura , 5., Asazuma, T., Takatori, T., "X-Ray study of the human neck motion due to head inertia loading" . Proceedings of the 38th Stapp Car Crash Conference, SAE Paper #942208, 55-64, 1994.

IRCOBI Conference - Hannover, September 1997 31

McElhaney, J .H . , Paver, J .G. , McCrackin, H .J . , Maxwel l , G .M . , "Cervical spine com­pression responses" . Proceedings of the 27th Stapp Car Crash Conference, SAE Paper #831615 , 163-177, 1983.

McElhaney, J . , Snyder, R.G. , States, J .D. , and Gabrielsen, M .A. , "Biomechanical a nalysis of swimming pool neck injuries" . Proceedings of the 23rd Stapp Car Crash Conference, SAE Paper #790137, 494-500, 1979.

Mertz, H . J . , Hodgson, V.R. , Thomas, L .M. , and Nyquist, G.W., "An assessment of com­pressive neck loads under injury-producing conditions" . Physician and Sportsmedicine, 6(11) :95-106, 1978.

Myers, B .S . , Richardson, W.J., Ooherty, B .J . , N ightingale, R.W. , McElhaney, J .H . , "The role of head constraint in the development of lower cervical compression flexion i njury" . Proceedings of the 35th Stapp Car Crash Conference, SAE Paper #912915, 391-399, 1991.

Myers, B .S . , Winkelstein, B.A. , "Epidemiology, classification, mechanism and toler­ance of human cervical spine injuries" . Critical Reviews in Biomedical Engineering 23(5&6) :307-409, 1995.

Myers, B .S . , N ightingale, R.W., "Basilar skul l fracture reslting from compressive neck loading" . A GARD Specialist Meeting on Impact Head lnjury, Mescalero, New Mexico, in press, November, 1996.

N ightingale, R.W., McElhaney, J .H . , Richardson, W.J . , Best, T.M. , Myers, B.S. , "Ex­perimental impact injury to the cervical spine: relating motion of the head and the mechanism of injury" . Journal of Bane Joint Surgery, 78-A(3): 312-321, 1996a.

N ightingale, R.W., McElhaney, J .H . , Richardson, W.J., Myers, B .S . , "Dynamic responses of the head a nd cervical spine to axial impact loading" . Journal of Biomechanics, 29(3): 307-318 , 1996b.

Nightingale, R.W., Richardson, W.J . , Myers, B .S . , 'The effects of padded surfaces on the risk for cervical spine injury" . Spine, (accepted) , 1997a.

Nightingale, R.W., McElhaney, J .H . , Camacho, D.L . , Winkelstein , B.A., Myers, B .S . , "The dynamic responses of the cervical spine: the role of buckling, end conditions, and tolerance in compressive impacts" . Proceedings of the 41th Stapp Car Crash Conference, 1997b.

N ightingale, R.W., Camacho, D.L . , Armstrong, A., Robinette, J . J . , Myers, B .S . , "Cer­vical spine flexibil ity and the effects of vertebral mass and loading rate on cervical spine buckling dynamics" . Journal of Biomechanics, (submitted) .

N usholtz, G .S . , Huelke, D .E . , Lux, P. , Alem, N .M . , Montalvo, F. , "Cervical spine in­jury mechanisms" . Proceedings of the 27th Stapp Car Crash Conference, SAE Paper #831616, 1983.

Panjabi , M .M„ White, A.A. 3d, Johnson, R .M. , "Cervical spine mechanics as a function of transection of components" . Journal of Biomechanics, 8(5):327-36, 1975.

Pintar, F.A., Sances, A. J r. , Yoganandan, N . , Reinartz, J. , Maiman, 0 .J . , Suh , J .K . , Unger, G . , Cusick, J.F. and Larson, S.J . , 'Biodynamics of the total human cadaveric cer­vical spine" . Proceedings of the 34th Stapp Car Crash Conference, SAE Paper #902309, 1990.

32 IRCOBI Conference - Hannover, September 1997

Pintar, F .A . , Yoganandan, N . , Pesigan, M. , Voo, L . , Cusick, J .F . , Maiman, D .J . , Sances, A. J r. , "Dynamic characteristics of the human cervical spine'.'._" Proceedings of the 39th Stapp Car Crash Conference, SAE Paper #95722, 195-202, 1995.

Roaf, R . , "A study of the biomechanics of spinal injuries" . Journal of Bone Joint Surgery, 428(4): 810-23, 1960.

Torg J .S . , Quendenfeld, T.C., Thieler, E.R. 1 1 1 , et a l . , "Collision with spring-loaded footba l l tackling and blocking dummies" . Journal of the American Medical Association, 236:1270-1271, 1976.

Torg J .S . , Vegso, J . J . , O'Nei l l , M .J . , Sennett, B., "The epidemiologic, pathologic, biomechanical, and cinematographic ana lysis of footba ll-induced cervical spine trauma" . American Journal of Sports Medicine, 18(1):50-57, 1990.

Walker, L .B .J r . , Harris, E.H. , Pontius, U .R . , "Mass, volume, center of mass, a nd mass moment of inertia of head and neck of the human body" . Proceedings of the 1 7th Stapp Car Crash Conference, 525-37, 1973.

Yoganandan, N . , Sances, A. Jr., Maiman, D.J . , Myklebust, J .B . , Pech, P. , and Larson, S .J . , "Experimental spinal injuries with vertical impact" . Spine, 1 1(9) :855-860, 1986.

fRCOBI Conference - Hamzover, September 1997 33